Correction of cognitive deficits in mouse models of Down ...1Institut de Géne ́tique et de...

RESEARCH ARTICLE

Correction of cognitive deficits in mouse models of Downsyndrome by a pharmacological inhibitor of DYRK1AThu Lan Nguyen1,2,3,4,5, Arnaud Duchon1,2,3,4, Antigoni Manousopoulou6, Nadeg̀e Loaëc5, Benoît Villiers5,Guillaume Pani1,2,3,4, Meltem Karatas7,8, Anna E. Mechling8, Laura-Adela Harsan7,8, Emmanuelle Limanton9,Jean-Pierre Bazureau9, François Carreaux9, Spiros D. Garbis6,*,‡, Laurent Meijer5,‡ and Yann Herault1,2,3,4,‡

ABSTRACTGrowing evidence supports the implication of DYRK1A in thedevelopment of cognitive deficits seen in Down syndrome (DS) andAlzheimer’s disease (AD). We here demonstrate that pharmacologicalinhibition of brain DYRK1A is able to correct recognition memorydeficits in three DS mouse models with increasing genetic complexity[Tg(Dyrk1a), Ts65Dn, Dp1Yey], all expressing anextra copyofDyrk1a.Overexpressed DYRK1A accumulates in the cytoplasm and at thesynapse. Treatment of the three DS models with the pharmacologicalDYRK1A inhibitor leucettine L41 leads to normalization of DYRK1Aactivity and corrects the novel object cognitive impairment observed inthese models. Brain functional magnetic resonance imaging revealsthat this cognitive improvement is paralleled by functional connectivityremodelling of core brain areas involved in learning/memoryprocesses. The impact of Dyrk1a trisomy and L41 treatmenton brain phosphoproteins was investigated by a quantitativephosphoproteomics method, revealing the implication of synaptic(synapsin 1) and cytoskeletal components involved in synapticresponse and axonal organization. These results encourage thedevelopment of DYRK1A inhibitors as drug candidates to treatcognitive deficits associated with DS and AD.

KEY WORDS: DYRK1A, Kinase inhibitor, Leucettine, Downsyndrome, Synapsin

INTRODUCTIONDown syndrome (DS) results from the trisomy of humanchromosome 21 (HSA21). It is still the most frequent intellectualdisability, affecting 1 newborn per 700 births. Among the mostcommon DS features are hypotonia, dysmorphic features andintellectual disability (Sureshbabu et al., 2011; Morris et al., 1982).Although children with DS show good socialization skills –encompassing social relations, friendship and leisure activities –they exhibit difficulties in communication abilities, i.e. the daily useof receptive, expressive and written language (Marchal et al., 2016).They experience troubles in daily life skills, such as self-caring,eating, toileting, dressing, behaving safely, and conceptualizingtime and money. Improving the intellectual quotient of DS peoplewould allow them to achieve more independence, increase theirvigilance and globally improve their quality of life.

Among candidate genes explaining intellectual disabilities in DSpeople, the dual specificity tyrosine-phosphorylation-regulatedkinase 1A, DYRK1A, is located in the DS chromosome 21 criticalregion (Walte et al., 2013; Duchon and Herault, 2016). It encodesa serine/threonine kinase which has numerous substrates. Twonuclear localization signals confer nuclear activity to this kinase(Alvarez et al., 2007), through interactions with transcription factorsincluding GLI1 (Mao et al., 2002), RNA POL II (Di Vona et al.,2015) or splicing factors like cyclin L2 (Graaf et al., 2004). In thecytoplasm, DYRK1A phosphorylates cytoskeletal substrates suchas β-tubulin, MAP1A or MAP1B (Ori-McKenney et al., 2016;Murakami et al., 2012; Scales et al., 2009). DYRK1A plays a role incell cycle regulation by phosphorylating the cyclin-dependentkinase (CDK) inhibitor KIP1 (also known as CDKN1B) in culturedhippocampal neurons and in embryonic mouse brain (Soppa et al.,2014) and LIN52 in vitro (Litovchick et al., 2011). Through its‘priming’ activity for glycogen synthase kinase 3β (GSK-3β)-dependent phosphorylation, DYRK1A regulates the nuclear/cytoplasmic localization of the NFAT transcription factors(Arron et al., 2006). At the synaptic level, DYRK1A binds toN-methyl-D-aspartate receptor subunit 2A (GLUN2A; also knownas GRIN2A) and synaptojanin 1 (SYNJ1) (Chen et al., 2014; Grauet al., 2014) and phosphorylates amphyphysin 1 (Murakami et al.,2012) and GLUN2A (Grau et al., 2014). These are examples ofdifferent biological brain functions controlled by DYRK1A whichare probably dysregulated when DYRK1A is overexpressed in DS,leading to cognitive impairments.

Several mouse models overexpressing DYRK1A have beendescribed. The first one, Tg(CEPHY152F7)12Hgc, carries a singlecopy of a yeast artificial chromosome (YAC) containing a 570 kbfragment of human DNA encompassing TTC3, DYRK1A andKCNJ6. This model shows no strong defect in spatial learning andmemory, but displays less crossing of the site where the platformwas during the probe test in the Morris water maze (MWM) taskReceived 11 May 2018; Accepted 1 August 2018

1Institut de Génétique et de Biologie Moléculaire et Cellulaire, Department ofTranslational Medicine and Neurogenetics, 67400 Illkirch, France. 2Centre Nationalde la Recherche Scientifique, UMR7104, 67400 Illkirch, France. 3Institut National dela Santé et de la Recherche Médicale, U964, 67400 Illkirch, France. 4Université deStrasbourg, 67400 Illkirch, France. 5ManRos Therapeutics, Perharidy ResearchCenter, 29680 Roscoff, Bretagne, France. 6Faculty of Medicine/Cancer Sciences &Clinical and Experimental Medicine, University of Southampton, Center forProteomic Research, Life Sciences Building 85, Highfield, Southampton SO17 1BJ,UK. 7Laboratory of Engineering, Informatics and Imaging (ICube), Integrativemultimodal imaging in healthcare (IMIS), UMR 7357, and University HospitalStrasbourg, Department of Biophysics and Nuclear Medicine, University ofStrasbourg, 67400 Illkirch, France. 8Department of Radiology, Medical Physics,Medical Center – University of Freiburg, Breisacher Strasse 60a, 79106 Freiburg,Germany. 9Université de Rennes 1, ISCR (Institut des sciences chimiques deRennes)-UMR, 6226, 35000 Rennes, France.*Present address: Proteome Exploration Laboratory, Division of Biology andBiological Engineering, Beckman Institute, California Institute of Technology,Pasadena, CA 91125, USA.

‡Authors for correspondence ([email protected];[email protected]; [email protected])

S.D.G., 0000-0002-1050-0805; L.M., 0000-0003-3511-4916; Y.H., 0000-0001-7049-6900

This is an Open Access article distributed under the terms of the Creative Commons AttributionLicense (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use,distribution and reproduction in any medium provided that the original work is properly attributed.

1

© 2018. Published by The Company of Biologists Ltd | Disease Models & Mechanisms (2018) 11, dmm035634. doi:10.1242/dmm.035634

Disea

seModels&Mechan

isms

mailto:[email protected]:[email protected]:[email protected]://orcid.org/0000-0002-1050-0805http://orcid.org/0000-0003-3511-4916http://orcid.org/0000-0001-7049-6900http://orcid.org/0000-0001-7049-6900http://creativecommons.org/licenses/by/3.0http://creativecommons.org/licenses/by/3.0

(Smith et al., 1997). Another model, Tg(MT1A-Dyrk1a)#Xest(#=9 or 33), was produced by expressing the Dyrk1a ratcomplementary DNA (cDNA) under the control of themetallothionein 1a exogenous promoter (Altafaj et al., 2001).These mice demonstrated impairments in neuromotor developmentand hyperactivity evaluated in treadmill performance and rotarodtests (Martínez de Lagrán et al., 2004). They also display defects invisuospatial learning and memory in the MWM task (Martínez deLagrán et al., 2004; Pons-Espinal et al., 2013), as well as inrecognition memory revealed in the novel object recognition (NOR)task (de la Torre et al., 2014). A third model, Tg(DYRK1A)36Wjs,was generated using a bacterial artificial chromosome (BAC)containing the human DYRK1A gene. DYRK1A triplication leads toalterations in synaptic transmission with an increase in long-termpotentiation (LTP) and a decrease in long-term depression (LTD).The transgenic mice are also deficient in the MWM task, suggestingspatial learning and memorization disabilities (Ahn et al., 2006).Although the human YAC and BAC transgenic mice exhibitfeatures similar to those seen in DS patients, they carry an extracopy of human/rat DYRK1A gene, which could lead to biasedphenotypes, as optimal expression and functionality of the human/rat protein cannot be ensured in a mouse background. Therefore, aBAC transgenic model with the entire Dyrk1a murine gene,Tg(Dyrk1a)189N3Yah [hereafter referred to as Tg(Dyrk1a)], wascreated (Guedj et al., 2012). This model shows alterations inshort-term memory in the Y-maze task, and in spatial memoryin the MWM task (Souchet et al., 2014). Deficits in corticalsynaptic plasticity were also observed (Thomazeau et al., 2014).Comparable impairments were seen in the Ts(1716)65Dn model(hereafter referred to as Ts65Dn), a mouse model trisomic foralmost 13.4 Mb, homologous to HSA21 and containing DYRK1A(Reeves et al., 1995). Spatial memory, especially reversal learningreflecting cognitive flexibility, was altered in the Water T-mazetest and in the reversal version of the MWM (Olmos-Serrano et al.,2016). Although the Ts65Dn model has been widely used tostudy DS features, it carries a triplication of genes located in asubcentromeric region of mouse chromosome 17 (MMU17) whichare not syntenic to any HSA21 genes (Duchon et al., 2011).A complete DS model, Dp1Yey, was thus produced, which istrisomic for 22.9 Mb, spanning the entire HSA21 region onMMU16 (Li et al., 2007). Dp1Yey mice are less well performingthan control mice in the MWM task and display context-associatedlearning deficits in the fear conditioning test (Yu et al., 2010).Reducing DYRK1A overdosage leads to correction of several DS

traits, demonstrating the major implication of this kinase in DS.Indeed, normalization of DYRK1A expression attenuates spatiallearning as well as associative memory defects, and rescues LTP inthe Ts65Dn model (García-Cerro et al., 2014; Altafaj et al., 2013).In addition, reversal to two DYRK1A copies in Dp1Yey miceenhances working and associative learning performance assessed inthe T-maze and contextual fear-conditioning tests, respectively(Jiang et al., 2015). Furthermore, epigallocatechin gallate (EGCG),a natural polyphenol found in coffee, cocoa and green tea, reportedto inhibit DYRK1A, restores intellectual capacities of trisomic mice(Guedj et al., 2009; de la Torre et al., 2014). EGCG has undergone aphase 2 clinical trial (de la Torre et al., 2016). However, EGCG alsointeracts with the cannabinoid receptor 1 (CNR1) (Korte et al.,2010). This receptor modulates the release of neurotransmitters invarious brain areas, such as the prefrontal cortex and hippocampus,thereby controlling memory, cognition processes and mood.Interaction of EGCG with CNR1 might thus affect memory,cognition and pain perception, leading to psychiatric disorders

(Freund et al., 2003; Wilson and Nicoll, 2002), compromising itstherapeutic use. Furthermore, DYRK1A is less sensitive to EGCG[half-maximal inhibitory concentration (IC50), 0.33 µM] thanvimentin (IC50, 0.003 µM) and the laminin receptor (IC50,0.04 µM) (Khan et al., 2006; Yang et al., 2009). Cognitiverestoration in trisomic mice by EGCG might thus be due toinhibition of targets other than DYRK1A. Consequently,alternative pharmacological inhibitors have started to emerge(Kim et al., 2016; Nakano-Kobayashi et al., 2017; Nguyen et al.,2017). Nevertheless, all available results clearly demonstratethe implication of DYRK1A in DS intellectual deficiencies andthe beneficial effects of its inhibition on the correction ofcognitive deficits.

DYRK1A has become a major screening target for thedevelopment of selective and potent pharmacological inhibitors(Smith et al., 2012; Stotani et al., 2016; Nguyen et al., 2017).We here investigated the effects of a relatively selective DYRK1Ainhibitor, leucettine L41 (hereafter referred to as L41) in threedifferent trisomic mouse models with increasing geneticcomplexity: Tg(Dyrk1a), Ts65Dn and Dp1Yey. Leucettines arederived from the marine sponge alkaloid Leucettamine B(Debdab et al., 2011; Tahtouh et al., 2012). The chemicallysynthesized L41 displays a high selectivity for DYRK1A but alsoDYRK1B, DYRK2 and some Cdc2-like kinases (CLKs) (Fig. 1).It acts by competing with ATP binding to the kinase catalytic site.We here establish a proof of concept that pharmacologicalinhibition of brain DYRK1A is able to correct NOR cognitiveimpairment in three DS models with increasing genetic complexity.We show, via brain functional magnetic resonance imaging (fMRI)in Dp1Yey, the most complete mouse model of DS, that suchcognitive improvement is paralleled by significant functionalconnectivity remodelling of core brain areas involved in learningandmemory processes. Furthermore, phosphoproteomic analyses inthe Tg(Dyrk1a) model unravelled brain DYRK1A targets for whichphosphorylation increases with DYRK1A overdosage anddecreases following L41 treatment. These novel substrates, suchas synapsin 1 (SYN1), also found in the phosphoproteomic analysesof Ts65Dn, the most used DS model, bring new insight into the roleof DYRK1A, and allow us to propose some dysregulated biologicalprocesses related to axonal organization and synaptic response whichare responsible for cognitive deficits associated with DS.

RESULTSLeucettines restore cognitive function, assessed in theNOR test, through kinase inhibition in three DS mousemodels overexpressing DYRK1ATo investigate the importance of DYRK1A in cognitive deficitsshown by transgenic mouse models of DS, we used a series of lowmolecular weight pharmacological inhibitors, collectively known asleucettines (Debdab et al., 2011; Tahtouh et al., 2012; T. Tahtouh,unpublished). We selected the well-characterized leucettine L41 asan archetype of this inhibitor family (Fig. 1) and L43, a closelyrelated analogue which displays little kinase inhibitory action.Because both compounds were found to inhibit CNR1 (T. Tahtouh,unpublished), we also used L99, a DYRK1A inhibitor lackingactivity on CNR1 (Fig. 1). To ensure its brain bioavailability,L41 was dosed following acute intraperitoneal (i.p.) injection inTg(Dyrk1a) and wild-type (wt) mice. Plasma half-life was∼45 min,and the inhibitor reached a maximum brain concentration at 20 min,and was eliminated 2 h later. No differences in L41 pharmacokineticsor biodistribution were observed between transgenic and wtmice (Fig. S1).

2

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035634. doi:10.1242/dmm.035634

Disea

seModels&Mechan

isms

http://dmm.biologists.org/lookup/doi/10.1242/dmm.035634.supplemental

We used three mouse models of DS: Tg(Dyrk1a), which expressesa single additional copy of DYRK1A (Guedj et al., 2012); andTs65Dn (Reeves et al., 1995) and Dp1Yey (Li et al., 2007), whichcarry MMU16 segments encompassing Dyrk1a, with 89 and 101genes homologous to HSA21, respectively (Gupta et al., 2016).Using the NOR test, we first evaluated the effects on Tg(Dyrk1a)

animals following daily i.p. treatment with L41 (20 mg/kg) for5, 12 or 19 days (Fig. 2A). As expected, untreated wt micediscriminated the novel over the familiar object. L41 treatment for5, 12 or 19 days had no effect on the performance of wt animals.Untreated Tg(Dyrk1a) mice were unable to discriminate the novelover the familiar object (de la Torre et al., 2014) (Fig. 2A). Incontrast, L41-treated Tg(Dyrk1a) mice preferentially explored thenovel object, thus reverting to the behaviour of wt animals. Thisrecovery was fully observed following 19 days of treatment, but wasconsistently or only marginally seen following 12 and 5 days oftreatment, respectively (Fig. 2A). In other words, a minimum of12 days of daily L41 treatment was necessary for full recovery in theNOR test.These experiments were repeated (daily i.p. treatment for

19 days) with the kinase-inactive/CNR1-active L43 and thekinase-active/CNR1-inactive L99 leucettines (Fig. 2B). Resultsclearly showed the beneficial behavioural effects of L99 (Fig. 2B,right) and the lack of effects of L43 (Fig. 2B, left), demonstratingthat the rescuing activity of leucettines derives from kinaseinhibition rather than CNR1 antagonism.We next ran the same experiments in Ts65Dn and Dp1Yey

animals (Fig. 2C). Daily i.p. treatment with L41 for 19 days led torescue in the NOR test. Intriguingly, L41 treatment had no restoringeffect on working memory (Fig. S2A), nor on place memory inTg(Dyrk1a) mice (Fig. S2B), as assessed in the Y-maze and placeobject location paradigms, respectively.

L41 treatment has a global effect on brain functionalconnectivity measured by resting state fMRITo noninvasively investigate whether DYRK1A kinase activityalters the brain functional connectivity (FC) and to reveal possiblecircuitry-based mechanisms underlying cognitive improvementsinduced by L41, we performed brain resting state fMRI (rsfMRI)experiments in vehicle or L41-treated Dp1Yey and wt mice.The brain connectivity patterns associated with default mode

network (DMN) – the main functional circuitry describing thebrain’s intrinsic activity at rest (Raichle, 2015) – were mappedcomparatively for each experimental group (Fig. 3A-a,b,B-a,b) viaseed-based analysis. The seed used for generating DMN was theretrosplenial cortex (RSP), considered as the mouse DMN core area.DMN configuration obtained for the wt vehicle-treated group(Fig. 3A-a) served as a control pattern and encompassed themidline cortical areas [RSP, posterior parietal association areas(PTLp), temporal association areas, visual areas] as well as therostral and medial anterior cingulate cortex (ACA) and hippocampalformation (HF) as previously described in mice (Sforazziniet al., 2014; Stafford et al., 2014). This DMN-like configurationwas only minimally impacted by L41 treatment in wt animals(Fig. 3A-c,d), by decreasing the RSP connectivity with limitedhippocampal (HF) areas.

In Dp1Yey mice, trisomy strongly influenced DMN architecture(Fig. 3B-a) by altering its pattern along midline cortical areas,highlighting the pathological features of Dp1Yey brain, as comparedwith wt brains (Fig. 3A-a). Notably, Dp1Yey mice show reversedconnectivity features of RSP (the core area of DMN) towards therostrofrontal cortical regions, including ACA [Fig. 3A-a versus B-a;switch from positive correlations (red/yellow scale) to negativecorrelations (blue scale)]. Intergroup statistics (vehicle-treated wtversus vehicle-treated Dp1Yey; Fig. S3) revealed diminishedRSP-ACA connectivity in Dp1Yey animals compared withcontrols (Fig. S3A), while strengthening the local connectivityaround the RSP seed (Fig. S3B). Concurrently, the RSP of Dp1Yeyvehicle animals showed increased connectivity to limbic areas ofbasal forebrain [i.e. pallidum (PAL)] when compared with that ofthe wt vehicle group (Fig. S3B).

L41 treatment of Dp1Yey mice rescued this altered DMNpattern (Fig. 3B-b), prominently acting to significantly increase theFC of the RSP with the ACA, prefrontal cortex (PFC) and ventralHF (group statistics in Fig. 3B-d, orange/red) and to reduce FCwith subcortical regions including the thalamus (TH) and PAL(Fig. 3B-d, green/blue).

To further reveal FC signatures of L41 action in Dp1Yey mice weevaluated the connectivity, across the whole brain, for several keybrain areas involved in learning and memory [hippocampal CA1and dentate gyrus (DG) areas, perirhinal cortex (PERI) and ACA].Group statistical analysis of FC maps highlighted overall restricted

Fig. 1. Chemical structure and selectivity of the leucettinesused in this study. Selectivity of leucettines L41, L43 and L99was assessed in vitro on 16 recombinant kinases, and in acellular CB1 annexin assay. Dose-response curves providedIC50 values (reported in µM). –, no inhibition at 10 µM.

3


Disea

seModels&Mechan

isms

http://dmm.biologists.org/lookup/doi/10.1242/dmm.035634.supplementalhttp://dmm.biologists.org/lookup/doi/10.1242/dmm.035634.supplementalhttp://dmm.biologists.org/lookup/doi/10.1242/dmm.035634.supplementalhttp://dmm.biologists.org/lookup/doi/10.1242/dmm.035634.supplementalhttp://dmm.biologists.org/lookup/doi/10.1242/dmm.035634.supplementalhttp://dmm.biologists.org/lookup/doi/10.1242/dmm.035634.supplemental

effects of L41 on brain FC of wt animals (Fig. 3C-a-d) but robustL41-dependent brain FCmodifications in theDSmodel (Fig. 3D-a-d).Acting at the hippocampal level, L41 treatment triggered robustchanges in CA1 and DG connectivity in Dp1Yey mice (Fig. 3D-a,b).The CA1 strengthened its FC with the PFC and ACA (Fig. 3D-a,orange/red) and decreased its functional communication with theventral HF (subiculum) and thalamic nuclei (Fig. 3D-a, green/blue).The strongest L41-triggered DG connectivity modifications wereidentified along the DG-RSP functional pathway in Dp1Yey mice(Fig. 3D-b). A divergent and limited effect of decreased CA1-ACAconnectivity was measured in wt mice, after L41 treatment(Fig. 3C-a, green/blue), and the DG altered its connectivity towardsthe TH and superior colliculus (SC) in wt animals.

Furthermore, L41 treatment triggered remodelling offunctional cross-talk between the PERI and the HF, RSP andPTLp in Dp1Yey animals (Fig. 3D-c), while acting primarilyon PERI-TH connectivity in wt animals (Fig. 3C-c). Groupstatistics additionally revealed a selective impact of L41 onACA connectivity in Dp1Yey mice (Fig. 3D-d), significantlymodifying its patterns towards the PFC (decrease), RSP(increase), SC (increase) and hypothalamic (HY) areas(decrease). Meanwhile, L41 induced limited effects in wt mice,by decreasing ACA-PFC connectivity (Fig. 3C-d). Overall, theseresults indicate the potential of L41 to act at a circuitry level,modifying the global brain FC in Dp1Yey mice, which arestrongly susceptible to its effects.

Fig. 2. DYRK1A-specific inhibitors rescue NOR deficits induced in Tg(Dyrk1a), Ts65Dn and Dp1yey trisomic mice. (A) Duration of treatment.NOR test results for Tg(Dyrk1a) mice treated with L41 or vehicle for 5, 12 or 19 days. Percentage object exploration by sniffing was determined for eachobject after a 24 h retention delay (familiar object, open symbol; novel object, filled symbol). NOR results of three Tg(Dyrk1a) cohorts treated with L41 for 5 (left),12 (centre) or 19 (right) days. Tg(Dyrk1a) and Ts65Dn treated animals spent more time exploring the novel object compared with control mice, showing a rescueof their recognition memory. Left: 5 days treatment induced a NOR rescue in Tg(Dyrk1a) animals [wt: n=15, P

Increased DYRK1A expression and catalytic activity in DSmodels: leucettines normalize DYRK1A activityTo validate the Tg(Dyrk1a) and Ts65Dn models in terms ofDYRK1A expression and function, we first verified the expressionlevels of Dyrk1a mRNA (Fig. 4A) and DYRK1A protein (Fig. 4B)in brains derived from control or L41-treated animals (19 days,daily i.p.). Total mRNAs were extracted from brains and Dyrk1a,

Gsk-3b and Rplp0 mRNAs were quantified by quantitativepolymerase chain reaction (qPCR) with specific primers. Resultsshowed the expected ∼1.5-fold increase in Dyrk1a mRNA levels(normalized with respect to Gsk-3b and Rplp0) in both transgenicmodels compared with their wt littermates. L41 treatment for19 days did not modify Dyrk1a mRNA levels (Fig. 4A). DYRK1Aprotein levels were also increased in transgenic mice models

Fig. 3. Influence of L41 on mouse brain functional connectivity (FC) patterns mapped via rsfMRI. (A,B) Default mode network (DMN) pattern in wt (a)and Dp1Yey (b) animals, mapped using the RSP cortex (core hub of DMN) as a seed region. A-a shows the typical DMN-like pattern observed in mice,spatially covering the middle rostrocaudal cortical axis of wt animals treated with vehicle, connecting the RSP and ACA. As shown in B-b, L41 treatment inwt animals slightly modifies the DMN patterns compared with wt-vehicle (see also statistics in A-c, sagittal view and A-d, coronal view; two-tailed Student’s t-test,P

compared with control wt animals, as shown by western blotting(WB) of total brain proteins, whereas GSK-3α/β and β-actin levelsremained at identical levels in transgenic and wt mice brains(Fig. 4B). L41 treatment had no effect on the expression ofDYRK1A and GSK-3α/β. We next measured DYRK1A catalyticactivities from transgenic and wt brain protein extracts (Fig. 4C).After 19 days of L41 or vehicle treatment, GSK-3α/β activityremained identical in the brains of transgenic and wt mice (data notshown), and was thus used to normalize the DYRK1A kinaseactivity. As expected, DYRK1A activity was elevated by ∼1.5- to1.8-fold in transgenic brains compared with wt brains (Fig. 4C).L41 treatment did not reduce DYRK1A activity in wt micebrains, but reduced DYRK1A activity by ∼30% in the brains ofTg(Dyrk1a) and Ts65Dn animals, essentially down to the level ofcontrol counterparts. DYRK1A kinase activity was thus normalizedby L41 treatment (Fig. 4C). In other words, although basal DYRK1Aactivity in trisomic and disomic mice brains was insensitive to L41,only excess DYRK1A activity in trisomic mice brains appeared tobe sensitive to L41. To verify that all brain DYRK1A activity can,in principle, be inhibited by L41, DYRK1A was extracted andimmunopurified from the brains of untreated wt and both transgenicanimals. DYRK1A kinase activities were assayed in vitro in thepresence of increasing concentrations of L41. Results showed that

the DYRK1A of wt and transgenic animal brains can be almostfully inhibited in vitro with essentially identical dose-responsecurves (Fig. 4D).

DYRK1A activity was measured following immunoprecipitation(and normalization on the basis of GSK-3α/β activity measured in thesame samples) from brain extracts of wt and Tg(Dyrk1a) animalstreated daily for 5, 12 or 19 days (Fig. 5A-C) with L41, or for 19 dayswith kinase-inactive L43 (Fig. 5D). As expected, DYRK1A activitywas increased in Tg(Dyrk1a) versus wt brains. Tg(Dyrk1a) brainDYRK1A activity was normalized after treatment with L41 for 12and 19 days, but not after 5 days of L41 treatment, nor after 19 days ofL43 treatment. These results correlate with L41-induced DYRK1Aactivity normalization (Fig. 5) and cognitive rescue (Fig. 2).

In all previous experiments, brains were collected 1 h after the lastleucettine treatment. We wondered about the persistence of theeffects of L41 after the last injection (Fig. 6). Tg(Dyrk1a) and wtanimals were treated with L41/vehicle daily for 19 days. NOR testswere run and brains collected 24 h or 48 h after the last L41treatment. DYRK1A catalytic activity was dosed in Tg(Dyrk1a) andwt mice brains. As expected, wt brain DYRK1A activity wasinsensitive to L41 treatment. Tg(Dyrk1a) brain DYRK1A activitywas increased compared with control wt brain DYRK1A activity(Fig. 6A,C), and normalized to wt levels 24 h after the last L41

Fig. 4. Dyrk1a mRNA and DYRK1A proteinexpression, and catalytic activity in Tg(Dyrk1a)and corresponding wt mice brains, and in Ts65Dnand corresponding wt mice brains. (A) mRNAexpression. Total RNA was extracted, purified andreverse transcribed into cDNA. mRNA expression ofDyrk1a, Gsk-3b and reference Rplp0 was quantifiedby qPCR from the amplification of cDNA with specificprimers (one primer annealing to an exon-exonjunction). Results are presented as mean±s.e. offour to six measurements and are shown relativeto Rplp0 expression, normalized to wt Gsk-3bexpression. (B) Protein expression. Total proteinswere extracted, resolved by SDS-PAGE and analysedby WB using antibodies directed against DYRK1A,GSK-3α/β and actin (loading control). (C) DYRK1Acatalytic activity. DYRK1A was purified from brainextracts by immunoprecipitation and GSK-3α/β waspurified by affinity chromatography on axin-agarosebeads. Activities of the purified kinases were assayedin triplicate in a radioactive kinase assay usingspecific peptide substrates, and are reported afternormalization with wt GSK-3α/β activities (mean±s.e.).(D) In vitro DYRK1A kinase activity. The catalyticactivity of DYRK1A immunoprecipitated from thebrains of Tg(Dyrk1a) and Ts65Dn mice and theirrespective controls was assayed in the presence of arange of L41 concentrations.

6


Disea

seModels&Mechan

isms

treatment (Fig. 6A). In contrast, L41 had no more effects 48 h afterthe last treatment (Fig. 6C). In terms of restoration of cognitiveabilities, the NOR tests revealed that Tg(Dyrk1a) deficits were stillcorrected 24 h, but not 48 h, after the last L41 treatment (Fig. 6B,D).Because L41 is essentially undetectable in brain extracts 2 h afterthe acute i.p. injection, it might be protected from degradation oncebound to DYRK1A or it could have been metabolized to anunidentified, stable active inhibitor.

Overexpressed DYRK1A accumulates in cytoplasm andsynapse: differential subcellular L41 distributionWe next investigated the subcellular distribution of DYRK1A in thebrains of Tg(Dyrk1a) and wt animals (Fig. 7A,B). Brains werecollected and cells dissociated and fractionated using two methods.The first allowed the separation of a cytosol+synaptosomesfraction from a nuclear fraction (Fig. 7A). The second separateda cytosol+nuclei fraction from a synaptosomal fraction (Fig. 7B).The purity of each fraction was evaluated by WB with specificmarkers: postsynaptic density protein 95 (PSD95; also known asDLG4) (cytosol+synaptosomes), histone H2B (nuclei), cyclin L1(cytosol+nuclei), SYN1 and AMPA-selective glutamate receptor1 (GLUR1; also known as GRIA1) (synaptosomal fraction)(Fig. 7A,B, top). DYRK1A expression levels were assessedfollowing sodium dodecyl sulfate–polyacrylamide gel electrophoresis(SDS-PAGE) of the different cellular fractions, followed by WB,and normalization to the levels of β-actin (Fig. 7A,B, bottom).DYRK1A was detected in all fractions in both genotypes, but itsexpression was significantly higher (∼1.5-fold), in the cytosol andsynaptosomes of Tg(Dyrk1a) brains compared with those of wtbrains. No differences in nuclear DYRK1A expression were seenbetween transgenic and wt animals. Brain DYRK1A overdosage inTg(Dyrk1a) animals thus occurs in the cytosol and synaptosomes,but not in the nuclei. We are currently exploring the reasons for thisdifferential distribution of excess DYRK1A.We next measured L41 levels in nuclear and cytoplasmic

fractions prepared from the brains of Tg(Dyrk1a) and wt animalswhich had been i.p. injected daily for 19 days with L41 (20 mg/kg)

or vehicle (Fig. 7C,D). At the end of the treatments, brains wererecovered and processed for L41 extraction and quantification byisobaric stable isotope chemical labelling, offline hydrophilicinteraction chromatography (HILIC), followed by ultra-highprecision liquid chromatography with electrospray ionization massspectrometry (LC–MS). Results show essentially undetectable L41 invehicle-treated animals, identical L41 levels in the brain nuclearfractions of Tg(Dyrk1a) and wt animals (Fig. 7C), and a significantlyincreased L41 level in the cytoplasmic fraction of Tg(Dyrk1a) brainscompared with the cytoplasmic fraction of wt animals’ brains(Fig. 7D). Thus, DYRK1A overexpression in the transgenic animals’brains appears to be limited to the cytoplasmic fraction,corresponding to the subcellular distribution of overexpressedDYRK1A (Fig. 7A,B). Accordingly, more L41 is detected in thecytoplasmic fraction from transgenic animals compared with theircontrol littermates.

Phosphoproteomic effects of DYRK1A trisomy and L41treatment reveal key synaptic and cytoskeletal componentsTo explore the mechanisms underlying the correcting effects ofL41 on NOR cognitive deficits of transgenic models, we analysedthe phosphoproteome of proteins isolated from the hippocampus,cortex and cerebellum of both Tg(Dyrk1a) and Ts65Dnmodels, along with their respective wt counterparts, and followingtreatment with vehicle or L41 (20 mg/kg, daily i.p. injection for19 days) (Fig. 8). All tissue samples were processed forphosphoproteomics analysis based on the enrichment andseparation of proteotypic phosphopeptides with HILIC (seeMaterials and Methods). In Tg(Dyrk1a) and Ts65Dn mice, thehippocampus, cortex and cerebellum yielded 1384, 1523 and 2004peptides, respectively, corresponding to 886, 948 and 1229 proteins(Table 1; Tables S1-S13).

Among the peptides/proteins detected in this study, only 30% ofthe proteins and 20% of the peptides were significantly up- ordownregulated in trisomic versus wt animals. Most peptides (80%)were phosphorylated on serine residues, whereas phosphorylationon threonine (15%) or tyrosine residues (5%) was less frequent.

Fig. 5. Effects of L41 treatment duration and treatmentwith L43. (A-D) Wt and Tg(Dyrk1a) mice were treatedwith L41 or vehicle for 5 (A), 12 (B) or 19 (C) days or L43 orvehicle for 19 days (D). Brains were recovered and extracted,and then DYRK1A and GSK-3α/β were immunopurified andaffinity purified, respectively, and assayed for their catalyticactivities. DYRK1A kinase activity was normalized withGSK-3α/β activities in each extract (mean±s.e.). DYRK1Ainhibition in Tg(Dyrk1a) mice brains was not significant after5 days of L41 treatment (P=0.42), but was increasinglysignificant after 12 (P=0.04) and 19 (P=0.01) days of L41treatment. 19 days treatment with kinase-inactive L43 did notreduce DYRK1A activity in Tg(Dyrk1a) mice (P=0.5).n.s., not significant. *P

Few peptides (less than 5%) were phosphorylated on two aminoacids. Very few phosphopeptides displayed the consensusDYRK1A phosphorylation sequence [R-P-x(1,3)-S/T-P] andmost phosphopeptides were predicted to be phosphorylated bykinases from the CMGC (MAPK or GSK-3 protein) or AGC[MTOR or PKG (also known as PRKG1)] groups (data obtainedwith the PhosphoRS algorithm within the Proteome Discoverersoftware tool, version 1.4).We selected the phosphopeptides displaying a trisomy-

associated modulation (up- or downregulation) which wasreverted by L41 treatment (down- or upregulation) (Fig. 8).These analyses were first run in each brain tissue and in each of thetwo models and their wt controls. We thus focused on proteinsdisplaying an L41-reversible, trisomy-associated phosphorylationmodulation. Based on these two criteria, 258 and 248phosphoproteins were selected from Tg(Dyrk1a) and Ts65Dnhippocampus (Fig. 8A), respectively. Similarly, the Tg(Dyrk1a)and Ts65Dn cortex showed 238 and 223 dysregulatedphosphoproteins, respectively (Fig. 8A). We found that 330 and341 phosphoproteins in Tg(Dyrk1a) and Ts65Dn cerebellum,respectively, were altered by trisomy and L41 treatment (Fig. 8A).Among these phosphoproteins, 102, 88 and 124 were common toboth transgenic models in the hippocampus, cortex

and cerebellum, respectively (Tables S1-S12). These sharedphosphoproteins were selected for DAVID cluster analysis(Tables S10-S12), which unravelled enrichment in synaptic,cytoskeletal and learning pathways (Fig. 8B; Fig. S4).ToppCluster analysis of the modulated phosphoproteins in eachmodel and each brain region confirmed enrichment in synaptictransmission common to both models in the hippocampus andcortex, while cytoskeleton organization was enriched in bothmodels for all three brain regions (Fig. 8B; Tables S11-S13).

We also compared, in each model, the phosphoproteinssubsets of all three brain areas (Fig. 8C). In Tg(Dyrk1a), only 16phosphoproteins were commonly modulated in the three brainsubstructures (Fig. 8C, left), while only 22 responded to thesecriteria in Ts65Dn (Fig. 8C, centre). Among these 16 and 22phosphoproteins shared by the three brain regions, only five werecommon to both DS models (Fig. 8D): the microtubule-associatedproteinsMAP1A,MAP1B andMAP2, and presynaptic componentspiccolo (PCLO) and SYN1. All phosphosites modulated byboth trisomy and L41 treatment, for each of the five proteins,are schematized in Fig. S5. They illustrate the complexity ofthe phosphoproteomics consequences of a single gene trisomy[Tg(Dyrk1a)] or a partial chromosome 16 trisomy (Ts65Dn)and the complexity resulting from the treatment with a single

Fig. 6. Persistence of the L41 inhibitory effect on DYRK1A activity and rescue of NOR deficit. (A,C) DYRK1A and GSK-3α/β activities were measured afterpurification from the brains of wt (n=3), L41-treated wt (n=3), Tg(Dyrk1a) (n=3) and L41-treated Tg(Dyrk1a) (n=3) animals, 24 h (A) or 48 h (C) following the end of a19 dayL41 treatment.After 24 h (P=0.02), but not at 48 h (P=0.93), theDYRK1Acatalyticactivityof the treatedTg(Dyrk1a)micebrainswasnormalizedcomparedwiththat of nontreated animals. (B,D) NOR tests were performed 24 h (B) or 48 h (D) after the last day of the 19 day L41 treatment. Although the rescuing effect wasdetectable 24 h after the last L41 treatment (P=0.01), no rescue was seen after a 48 h delay (P=0.72). n.s., not significant. *P

pharmacological agent. Among these five proteins, we looked forthe residues with increased phosphorylation when DYRK1A wasoverexpressed, and with reduced phosphorylation when DYRK1Awas inhibited by L41, and also matching the consensus DYRK1Aphosphorylation sequence (Himpel et al., 2000). Based on thesecriteria, serine 551 of SYN1 was selected for further study.

DYRK1A interacts with SYN1 and other proteins implicatedin synaptic functionsTo investigate potential interactions between DYRK1A andSYN1, co-immunoprecipitation (co-IP) experiments were carriedout with adult mouse brain lysates (Fig. 9A) using antibodiesdirected against SYN1 or DYRK1A (negative control, GAPDH).As expected, DYRK1A and SYN1 were found in their respectiveimmunoprecipitates (IPs). SYN1 was detected in DYRK1A IPsand DYRK1Awas detected in SYN1 IPs (Fig. 9A), suggesting thatthese proteins form a direct or indirect complex in brain extracts.Calmodulin-dependent kinase 2A (CAMK2A) was present inSYN1 IPs, as expected from previous results (Llinás et al., 1985;Benfenati et al., 1992) and from its role in presynaptic vesicle pool

release (Cesca et al., 2010). CAMK2A was also detected inDYRK1A IPs, suggesting the possibility of a DYRK1A/SYN1/CAMK2A complex, although separate DYRK1A/CAMK2A andSYN1/CAMK2A complexes are possible.

To see whether DYRK1A directly phosphorylates SYN1, we ranin vitro kinase assays using recombinant DYRK1A and variousSYN1-derived peptides, including Ser551, as potential substrates, orWoodtide as a reference substrate (Fig. 9B,C). Recombinant DYRK1Adisplayed similar activity towards SYN1-tide or SYN1-S553A-tidecompared with Woodtide. In contrast, no significant phosphorylationcould be measured with the SYN1-S551A peptide. This confirms thatDYRK1A is able to phosphorylate SYN1 on its S551 residue, but noton the nearby Ser553 site. The Ser551 site matches with the consensusDYRK1A phosphorylation site (Fig. 9B).

DISCUSSIONRescue of cognitive deficits by pharmacological inhibitionof excess DYRK1AIn this study, we show that trisomy is associated with an increase inDYRK1A expression and catalytic activity, and that a class of

Fig. 7. DYRK1A and L41 subcellular localization. (A,B) Wt or Tg(Dyrk1a) brains were fractionated by two methods and the expression of DYRKA wasestimated by WB following SDS-PAGE. Reference subcompartment-specific proteins were detected by WB. (A) DYRK1A expression in cytoplasm+synaptosomes and in nuclear fractions [wt, n=6; Tg(Dyrk1a), n=6]. DYRK1A overexpression is observed in the cytoplasm+synaptosomes fraction (P=0.006),but not in the nuclear fraction (P=0.9). WB of specific markers validates the purity of fractions: PSD95 (95 kDa, cytoplasmic+synaptosomal marker), H2B(17 kDa, nuclear marker), β-actin (42 kDa, housekeeping protein). (B) DYRK1A expression in cytoplasm+nuclei and in synaptosomal fractions [wt, n=7;Tg(Dyrk1a), n=7]. DYRK1A was overexpressed in both cytoplasmic+nuclear (P=0.001) and synaptosomal (P=0.02) fractions. Fractionation was confirmedby WB of specific compartment markers: cyclin L1 (55 kDa, cytoplasmic+nuclear marker), GLUR1 (100 kDa, postsynaptic marker), SYN1 (74 kDa, presynapticmarker), β-actin (42 kDa, housekeeping protein). (C,D) L41 subcellular levels. L41 was more highly detected in the brain nuclear (C) and cytoplasmic(D) compartments in L41-treated wt (n=5) and Tg(Dyrk1a) (n=2) mice compared with nontreated wt (n=5) and nontreated Tg(Dyrk1a) (n=5) mice. L41 distributionwas not significantly different between the brain nuclear fractions of treated wt and Tg mice. In contrast, the L41 level was increased in the cytoplasm oftreated Tg mice brain compared with the cytoplasm of control wt mice brain (P=0.002). n.s., not significant. *P

synthetic DYRK1A inhibitors, the leucettines, exemplified by L41,is able to cross the blood brain barrier and selectively inhibitthe excess DYRK1A linked to trisomy. Why only this fractionof overexpressed DYRK1A is inhibited, and most native, basalDYRK1A is not, remains a mystery. This effect could be linked tothe accumulation of excess DYRK1A and L41 in specific cellularcompartments and not in others, as shown in Fig. 7. Intriguingly,a similar sensitivity to inhibition of excess DYRK1A comparedwith ‘normal’ DYRK1A was observed with EGCG. This finding isencouraging in terms of potential therapeutic implications, ascomplete inhibition of DYRK1A is not desired. IntracellularDYRK1A has been described in both nuclear and cytoplasmcompartments (Martí et al., 2003). Our results indicate that it is alsopresent in synaptosomes, which might have consequences on theregulation of synaptic vesicles trafficking (see below).We here demonstrate the rescuing effect of synthetic DYRK1A

inhibitors, leucettines L41 and L99, on deficient recognition

memory of three different trisomic mouse models with increasinggenetic complexity, Tg(Dyrk1a), Ts65Dn and Dp1Yey. Thesebeneficial behavioural effects directly correlate with inhibitionof excess DYRK1A activity. There is also a strong coincidencewith the duration of the drug treatment (Figs 2A and 5), thepotency of the leucettine analogues (Figs 2B and 5) andthe duration of the drug-free period following the last injection(Fig. 6). Finally, behavioural correcting benefits detected in theNOR test (Figs 2 and 6) correlate with remodelling of brainfunctional connectivity detected by fMRI (Fig. 3). However, weobserved that working and spatial memories impaired in theTg(Dyrk1a) mice were insensitive to L41 treatment, as assessed inthe Y-maze and place object location tasks, respectively. Thisindicates a specific action of DYRK1A inhibition on molecularpathways specifically related to recognition memory. Our findingsfurther strengthen the essential role of DYRK1A in intellectualphenotypes associated with DS. Leucettine derivatives should thus

Fig. 8. Phosphoproteomic analysis of Tg(Dyrk1a) and Ts65Dn mice brains following exposure to L41. Phosphoproteins, in each brain substructure,that are both up- or downregulated by trisomy and, respectively, down- and upregulated by L41 treatment were selected for analysis. (A) Venn diagramscomparing the two transgenic models versus wt and L41 treatment, at tissue level. 102, 88 and 124 modified phosphoproteins were common to the two modelsin the hippocampus, cortex and cerebellum, respectively. Numbers in parentheses indicate dual modulated phosphoproteins in each model and each tissue.(B) Biological processes enrichment deregulated by the phosphoproteins which are modulated one way in both Tg(Dyrk1a) and Ts65Dn mice and affected byL41 treatment in the opposite way. Represented here are those common to both models and to the three brain tissues. DAVID and ToppCluster analyseswere performed in the hippocampus, cortex and cerebellum separately. Enriched classification is determined by the –log(P-value). Synaptic transmission,common to the hippocampus and cortex, and cytoskeleton organization, common to the three brain regions, are the processes most modified by trisomy andsensitive to L41. (C) Venn diagrams illustrate the number of dually modulated phosphoproteins in each model and in each tissue, and the numbers shared bydifferent brain areas. 16 and 22 proteins were shared by all three tissues in Tg(Dyrk1a) and Ts65Dnmice, respectively. (D) Venn diagram comparison of these 16and 22 phosphoproteins revealed that five are shared by both models.

10


Disea

seModels&Mechan

isms

be investigated further as drug candidates to improve cognitivefunctions of DS patients.

L41 treatment in DYRK1A-overexpressing mice triggersremodelling of brain FC pathwaysBrain rsfMRI in Dp1Yey mice revealed global resilienceof functional cerebral circuitry after L41 administration.Notably, L41 corrected the abnormal DMN patterns found in

Dp1Yey mice, but also acted on connectivity of key brain areasassociated with cognitive and memory processing (PFC, ACA,PERI, HF). DMN (Raichle, 2015) – previously described as ahighly active circuitry during rest – and preserved across species(Stafford et al., 2014), was shown to be vulnerable to variousneuropathological conditions (Hawellek et al., 2011; Raichle, 2015;Zhou et al., 2017), including DS (Anderson et al., 2013; Pujol et al.,2015). The core area of this network in mice is the RSP (associated

Table 1. Summary of phosphoproteomic analyses

ModelTg(Dyrk1a) Ts65Dn Common

Brain area Hippocampus Cortex Cerebellum Hippocampus Cortex Cerebellum Hippocampus Cortex Cerebellum

Total protein number 886 948 1229 886 948 1229 – – –Total peptide number 1384 1523 2004 1384 1523 2004 – – –Modulated proteins, trisomic vs wt 275 256 364 257 230 365 – – –Modulated peptides, trisomic vs wt 333 296 437 311 270 422 – – –Modulated proteins, trisomic, L41 vs vehicle 267 240 344 253 228 355 34 38 59Modulated peptides trisomic, L41 vs vehicle 321 275 403 307 265 410 40 38 63Phospho-Ser peptides 265 221 333 246 205 320 33 29 51Phospho-Thr peptides 39 44 58 53 42 75 6 6 11Phospho-Tyr peptides 17 10 12 8 18 15 1 3 1Phospho-Ser and phospho-Thr peptides 18 19 23 10 6 24 – – –Phospho-Ser and phospho-Tyr peptides 5 1 4 2 3 4 – – –Phospho-Thr and phospho-Tyr peptides 6 1 4 7 4 2 – – –DYRK1A phosphorylation sites 1 7 8 6 3 6 – – –Other kinases’ phosphorylation sites 279 251 383 263 223 351 – – –

Fig. 9. Direct interaction of DYRK1A and SYN1, phosphorylation of SYN1 by DYRK1A. (A) DYRK1A and SYN1 were immunoblotted followingimmunoprecipitation from wt mice brain extracts. DYRK1A or SYN1 present in the starting material (Input) were recovered in the IPs. SYN1 (74 kDa) was presentin the DYRK1A IP and DYRK1A (85 kDa) was detected in the SYN1 IP, suggesting that these two proteins interact directly. Positive control of the SYN1 IP wasperformed using an anti-CAMKII antibody. As expected, CAMKII (50 kDa) was present in the SYN1 IP. DYRK1A IP also brought down CAMKII, suggestingcomplexes between SYN1, CAMKII and DYRK1A. (B) Sequence of SYN1 in the vicinity of Ser551 matches with the consensus DYRK1A phosphorylation site.Based on this sequence, three peptides were synthesized and used as potential substrates: SYN1, SYN1-S551A and SYN1-S553A. (C) Kinase activity ofrecombinant DYRK1A towards the three different SYN1peptides. SYN1 and SYN1-S553A peptides were phosphorylated at the same level as Woodtide byrecombinant DYRK1A (71.7%±5.2%, 70.1% and 78.4%±11.4%, respectively). No significant catalytic activity was measured with the SYN1-S551A peptide(7.9%±1.2%).

11


Disea

seModels&Mechan

isms

with the posterior cingulate/precuneus cortex in humans) (Hübneret al., 2017; Sforazzini et al., 2014). Our analysis unravelledincreased local connectivity around the RSP in DS mice, butstrongly reduced long-range communication with frontocorticalbrain regions (ACA, PFC; Fig. S3), when compared with wtanimals. This short-range stronger connectivity is not limited tothe RSP, but represented a common feature for other investigatedbrain regions (ACA, PERI, HF) of Dp1Yey mice. Such a pattern ofincreased local, short-range brain communication was describedas a cardinal feature of FC in DS patients (Anderson et al., 2013;Pujol et al., 2015; Vega et al., 2015). Indeed, DS human brains arecharacterized by simplified network structure, organized by localconnectivity (Anderson et al., 2013; Pujol et al., 2015; Vega et al.,2015) and impaired efficiency to integrate information fromdistant connections.Dp1Yey mouse brains additionally displayed features of higher

negative functional correlations as compared with the wt vehiclegroup and, more obviously, a reversed correlation pattern (switchfrom positive to negative correlations) between the RSP and frontalcortical areas in the DS model. This feature, attenuated or correctedfollowing L41 treatment, could eventually be discussed in the contextof L41 regulation of inhibition/excitation ratios, imbalanced inDYRK1A-overexpressing mice (Souchet et al., 2014). Indeed,increased number and signal intensity from neurons expressingGAD67 (also known as GAD1), an enzyme that synthesizes GABA,indicating inhibition pathway alterations, was quantified in threedifferent DS models (Souchet et al., 2014), including Dp1Yey.Pharmacological correction of inhibition/excitation was achieved inthe Tg(Dyrk1a) DS mouse model (Souchet et al., 2015) by EGCGtreatment. We can speculate on a similar effect of L41 on inhibition/excitation balance, and subsequent modulation of brain connectivity.Nevertheless, the brain synchrony modifications after L41 inhibitionof excess DYRK1A activity in DS models might potentially reflectother molecular mechanisms and interactions at the synaptic andcytoskeletal level, as shown here, and subsequently underpincorrection of cognitive and memory deficits of DS mice.Importantly, L41 had only limited effects on FC in wt animals,whereas in the Dp1Yey model it largely impacted the connectivityfeatures, on distributed action sites, that coincide with alterationsreported for brain anatomy in DS models, most notably, frontaland prefrontal cortical areas (ACA/PFA), the HF, PAL and TH.Volumetric MRI in DS mouse models, showed a general trend forsmaller frontal lobes, hippocampal and cerebellar regions, butlarger thalamic and hypothalamic areas (Powell et al., 2016;Roubertoux et al., 2017). Diffusion MRI also identified potentialmicrostructural alterations in the above-mentioned areas and alsothe striatum (including the PAL) (Nie et al., 2015). Our rsfMRIstudy advances the current knowledge on the brain functionalcommunication in DSmousemodels, revealing targeted and effectiveaction of L41 on brain circuitry, consistent with the profile ofcognitive and novel object recognition memory improvements.

DYRK1A and SYN1Phosphoproteomic analyses using ultra-high precision LC–MSanalysis unravelled several clusters of neuronal phosphorylatedproteins directly controlled by DYRK1A or clusters indirectlymodulated in the trisomic condition and sensitive to L41 treatment.Five phosphoproteins were shared by Tg(Dyrk1a) and Ts65Dnmice and were present in three brain substructures (hippocampus,cortex, cerebellum) (Fig. 8). Furthermore, these phosphoproteinsshowed significant modulation in their phosphorylation levels intrisomic versus disomic animals and these modulations were

sensitive, in the opposite direction, to L41 treatment. A few keypathways, including controlling synaptic vesicle (SV) transport,calcineurin NFAT signalling and cytoskeleton organization, werefound to be directly affected by DYRK1A, or as a consequence of itskinase activity (Fig. S4), while others might represent indirecteffects of the overdosage. Nevertheless, the immune response wasfound to be affected, correlating with several studies linkingDYRK1A to inflammation. We here focused on SYN1 as it was theonly protein that revealed a serine residue corresponding to theDYRK1A phosphorylation consensus sequence. SYN1 Ser551was hyperphosphorylated following DYRK1A overexpression anddephosphorylated following L41 treatment. Representative annotatedultra-high resolution product ion spectrum of proteotypic peptideqSRPVAGGPGAPPAARPPAsPSPqR encoding the phosphorylatedresidue Ser551 is shown in Fig. S6.

Co-IP experiments showed that DYRK1A interacts, either directlyor indirectly, with SYN1 (Fig. 9). SYN1 has been described to beinvolved in the reserve SV pool maintenance at the presynapticbouton by tethering SVs to the actin cytoskeleton (Hilfiker et al.,2005; Benfenati et al., 1991). Phosphorylation of SYN1 by CAMKIIleads to the release of SVs and allows them tomove close to the activezone (Llinás et al., 1991). Neurotransmitter release at the active zoneis thus strongly dependent on SYN1 phosphorylation. We showedthat SYN1 was phosphorylated by DYRK1A on its S551 residuein vitro and in vivo, thus highlighting a novel role of DYRK1Ain SYN1-dependent presynaptic vesicle trafficking. Besides itsphysiological role in synaptic plasticity regulation, SYN1 has beenassociated with epilepsy (Garcia et al., 2004; Fassio et al., 2011).Mutations in the phosphorylation domains of SYN1 essential forvesicle recycling control have been related to epilepsy (Fassioet al., 2011). In addition, mental retardation, autosomal dominant7 (MRD7) patients with Dyrk1a haploinsufficiency display epilepsyseizures (Courcet et al., 2012; Møller et al., 2008; Oegema et al.,2010; Valetto et al., 2012; Yamamoto et al., 2011). Our resultssuggest that epileptic seizures observed in MRD7 patients could beinduced by defects in SYN1 regulation.

DYRK1A, microtubule-binding proteins, PCLO and othersynaptic targetsThe other four proteins found in our phosphoproteomics study willbe the object of another study but are briefly reviewed here. Thedetection of MAP1A, MAP1B and MAP2, all previously reportedas DYRK1A substrates (Murakami et al., 2012; Scales et al.,2009), validated the power of our analysis and confirmed therole of DYRK1A in dendrite morphogenesis and microtubuleregulation (Ori-McKenney et al., 2016). The last phosphoprotein,PCLO, is a cytoskeletal matrix protein associated with thepresynaptic active zone (Cases-Langhoff et al., 1996), whichacts as a scaffolding protein implicated in SV endocytosisand exocytosis (Garner et al., 2000; Fenster et al., 2003). Thelack of PCLO in the human brain leads to a dramatic neuronalloss associated with pontocerebellar hypoplasia type III (Ahmedet al., 2015). Moreover, PCLO knockdown in culturedhippocampal neurons increases SYN1 dispersion out of thepresynaptic terminal and SV exocytosis (Leal-Ortiz et al., 2008).It has been shown that PCLO modulates neurotransmitter release byregulating F-actin assembly (Waites et al., 2011). It clearly appearsthat PCLO acts upstream of SYN1 and regulates its role in vesiclerecycling.

Taken together, our findings reveal SYN1 as a new directsubstrate of DYRK1A, suggesting a novel role of this kinase in theregulation of SV release at the presynaptic terminal. Moreover, the

12


Disea

seModels&Mechan

isms

http://dmm.biologists.org/lookup/doi/10.1242/dmm.035634.supplementalhttp://dmm.biologists.org/lookup/doi/10.1242/dmm.035634.supplementalhttp://dmm.biologists.org/lookup/doi/10.1242/dmm.035634.supplemental

relatively safe and selective DYRK1A inhibitors, the leucettines,successfully correct recognition memory deficits associated withDS in three different mice models. Although the DYRK1A-dependent biological process which is rescued by these drugs stillneeds to be elucidated, leucettines and their analogues representpromising therapeutic drugs to enhance cognitive functions in DSpatients.

DYRK1A, DS and ADThere is strong support for the involvement of DYRK1Ain cognitive deficits associated with Alzheimer’s disease (AD):(1) DYRK1A mRNA (Kimura et al., 2007) and DYRK1A protein(Ferrer et al., 2005) levels are increased in postmortem human ADbrains compared with healthy brains; (2) calpain 1-induced cleavageof DYRK1A is observed in AD brains and associated with increasedactivity (Jin et al., 2015); (3) DYRK1A phosphorylates key ADplayers, such as amyloid precursor protein (Ryoo et al., 2008),presenilin 1 (Ryu et al., 2010), Tau (also known as MAPT) (Woodset al., 2001; Ryoo et al., 2007; Azorsa et al., 2010; Coutadeur et al.,2015; Jin et al., 2015), septin (Sitz et al., 2008) and neprylysin(Kawakubo et al., 2017); (4) DYRK1A regulates splicing ofTau mRNA (Shi et al., 2008; Wegiel et al., 2011; Yin et al., 2012,2017; Jin et al., 2015); (5) DYRK1A inhibition corrects cognitivedefects in 3xTG-AD (Branca et al., 2017), APP/PS1 (B. Souchet,unpublished) and Aβ25-35 peptide-injected wt mice (Naert et al.,2015), three widely used mice models of AD. These facts provideadditional incentive to investigate the regulation and substratesof brain DYRK1A and to develop potent and selective DYRK1Ainhibitors to treat cognitive deficits observed in differentindications. DS patients display early symptoms of AD and showa high frequency of dementia at later age (Ballard et al., 2016). Thetriplication of APP located on the HSA21 is thought to contribute toamyloid plaques and neurofibrillary tangles, two causative factorsin AD, that accumulate early in 30- to 40-year-old DS people (Head,et al., 2012a). These factors, associated with neuroinflammation andoxidative damage also diagnosed in both AD and DS individuals,lead to precocious dementia observed from age 30 to 39 (Head et al.,2012b). Studying DS will have an impact on the understanding ofAD and, reciprocally, DYRK1A is clearly a common factor betweenthe two diseases.

MATERIALS AND METHODSAnimal models, treatment and behaviour assessmentTg(Dyrk1a) mutant mice and Dp1Yey models were maintained on theC57BL/6J genetic background. Ts(1716)65Dn trisomic mice were obtainedfrom The Jackson Laboratory and kept on the C57BL/6J×C3B, a congenicsighted line for the BALB/c allele at the Pde6b locus (Hoelter et al., 2008).The local ethics committee, Com’Eth (no. 17), approved all mouseexperimental procedures, under the accreditation number APAFIS #5331and #3473, with Y.H. as the principal investigator.

Behavioural studies were conducted in 12- to 20-week-old male animals.All assessments were scored blind to genotype and treatment, asrecommended by the ARRIVE guidelines (Karp et al., 2015; Kilkennyet al., 2010). Leucettine L41 was prepared at 40 mg/ml in dimethylsulfoxide (DMSO), aliquoted and stored below −20°C. The final formulationwas prepared just prior to use as a 2 mg/ml solution diluted in PEG300/water(50/45), to reach a final DMSO/PEG300/water 5/50/45 (v/v/v) mix.Treated animals received a daily dose (5, 12 or 19 days) of this formulationby i.p. injection of 20 mg/kg/day. Nontreated animals received the sameformulation without L41.

The NOR task is based on the innate tendency of rodents to differentiallyexplore a novel object over a familiar one (Ennaceur and Delacour, 1988).Day 1 was a habituation session. Mice freely explored the apparatus, a whitecircular arena (53 cm diameter) placed in a dimly lit testing room (40 lux).

On day 2, the acquisition phase, mice were free to explore two identicalobjects for 10 min. Mice were then returned to their home cage for a 24 hretention interval. To test their memory, on day 3, one familiar object(already explored during the acquisition phase) and one novel object wereplaced in the apparatus and mice were free to explore the two objects for a10 min period. Between trials and subjects, the different objects werecleaned with 50° ethanol to reduce olfactory cues. To avoid a preference forone of the objects, the new object was different for different animal groupsand counterbalanced between genotype and treatment as well as for locationof novel and familiar objects (left or right). Object exploration was manuallyscored and defined as the orientation of the nose to the object at a distance

5 mM EDTA, 5 mM EGTA, 0.1% Nonidet P-40 and protease inhibitorcocktail from Roche, France), were added to the mix and gently rotatedat 4°C for 30 min. After a 1 min spin at 10,000 g and removal of thesupernatant, the pelleted immune complexes were washed three times withbead buffer, and a last time with Buffer C (60 mM β-glycerophosphate,30 mM p-nitrophenolphosphate, 25 mM Mops pH 7.2, 5 mM EGTA,15 mM MgCl2, 2 mM dithiothreitol, 0.1 mM sodium orthovanadate,1 mM phenylphosphate, protease inhibitor cocktail). DYRK1A or GSK-3immobilized on beads were assayed in buffer C as described in theSupplementaryMaterials andMethods withWoodtide (KKISGRLSPIMTEQ)(1.5 µg/assay) or GSK3-tide (YRRAAVPPSPSLSRHSSPHQpSED-EEE,where pS stands for phosphorylated serine) as substrates.

Protein kinase assays to evaluate SYN1 phosphorylation by DYRK1Awere performed with 50 ng recombinant DYRK1A protein (PV3785,Thermo Fisher Scientific) and 0.98 mM Woodtide, and three peptidesderived from the SYN1 putative DYRK1A phosphorylation site (Fig. 9C).Kinase activity was then measured as described in the SupplementaryMaterials and Methods.

The selectivity of the three leucettines used in this study was evaluatedin a panel of 16 recombinant protein kinases assayed as described in theSupplementary Materials and Methods.

Subcellular fractionationNuclear, cytosolic and synaptosomal subcellular fractionation of braintissue was performed with the Syn-PER™ and ProteoExtract® TissueDissociation Buffer Kit and Subcellular Proteome Extraction Kit followingthe instructions of the manufacturer. Fractions were analysed by SDS-PAGEand WB with specific antibodies.

Phosphoproteomics results analysisGene ontology enrichment analyses of phosphoproteins that are modulated(up- or downregulated) in Tg(Dyrk1a) or Ts65Dn mice versus wt and alsomodulated in the opposite manner (down- or upregulated) by the L41treatment, were conducted using ToppCluster (Bonferroni correction,P-value cutoff 0.05). Only biological processes common to the threebrain regions and both models are presented (complete biological processesare listed in Tables S11-S13).

DYRK1A substrates and their respective phosphorylation sites wereidentified in the phosphoproteome based on the DYRK1A phosphorylationconsensus sequence R-P-x(1,3)-S/T-P (Himpel et al., 2000). Protein-proteininteractions of each substrate were generated with STRING web serverapplication. Biological process enrichments of each cluster were assessed byusing ToppCluster web server application. Phospho-network was mappedwith the Cytoscape tool. See Supplementary Materials and Methodsfor details.

Immunoprecipitation and immunoblottingAll immunoprecipitations were performed on fresh half brains of 3-month-old wt male mice. Brains were dissected and lysed in 1.2 ml RIPA lysisbuffer (Santa-Cruz Biotechnology, France) using Precellys® homogenizertubes. After centrifugation at 2800 g for 2×15 s, 1 ml brain extract wasincubated with 2 µg of antibody of interest at 4°C for 1 h under gentlerotation. An aliquot of the remaining supernatant was kept for furtherimmunoblotting as homogenate control. Then, 20 µl protein G agarosebeads, previously washed three times with bead buffer, were added to themix and gently rotated at 4°C for 30 min. After a 1 min spin at 10,000 g andremoval of the supernatant, the pelleted immune complexes were washedthree times with bead buffer beforeWB analysis with appropriate antibodiesdirected against DYRK1A (H00001859 M01, Interchim; 1:1000), PSD95(ab18258, Abcam, France; 1:1000), SYN1 (ab64581, Abcam; 1:1000),CAMK2A (PA5-14315, Thermo Fisher Scientific; 1:1000) and GAPDH(MA5-15738, Thermo Fisher Scientific; 1:3000). Immunoblots wererevealed with Clarity Western ECL Substrate (Bio-Rad).

Competing interestsL.M. is founder, CEO and CSO of ManRos Therapeutics, which licensed thepatent on leucettines and develops these as DS/AD drug candidates. L.M., F.C. andJ.-P.B. are co-inventors on the leucettine patent.

Author contributionsConceptualization: T.L.N., A.D., S.D.G., L.M., Y.H.; Methodology: T.L.N., A.D., N.L.,B.V., M.K., A.E.M., L.-A.H., E.L., J.-P.B., F.C., S.D.G.; Software: M.K., A.E.M.,L.-A.H., S.D.G.; Validation: T.L.N., B.V., L.M.; Formal analysis: T.L.N., A.M., M.K.,A.E.M., L.-A.H., S.D.G.; Investigation: T.L.N., A.D., A.M., N.L., B.V., G.P., M.K.,A.E.M., J.-P.B., F.C., S.D.G.; Resources: A.M., E.L., J.-P.B., F.C.; Data curation:A.M., M.K., A.E.M., L.-A.H., S.D.G.; Writing - original draft: T.L.N., L.M.; Writing -review & editing: T.L.N., S.D.G., L.M., Y.H.; Supervision: S.D.G., L.M., Y.H.; Projectadministration: L.M., Y.H.; Funding acquisition: L.M., Y.H.

FundingThis work was supported by Fonds Unique Interministériel (TRIAD project;L.M., Y.H., J.-P.B.), Conseil Régional de Bretagne (L.M., Y.H., J.-P.B.),Fondation Jérôme Lejeune (L.M.), Seventh Framework Programme (BlueGenics;L.M.), Agence Nationale de la Recherche (Programme Investissements d’Avenir;ANR-10-IDEX-0002-02, ANR-10-LABX-0030-INRT, ANR-10-INBS-07 PHENOMINto Y.H.) and CIFRE (T.L.N.).

Supplementary informationSupplementary information available online athttp://dmm.biologists.org/lookup/doi/10.1242/dmm.035634.supplemental

ReferencesAhmed, M. Y., Chioza, B. A., Rajab, A., Schmitz-Abe, K., Al-Khayat, A.,

Al-Turki, S., Baple, E. L., Patton, M. A., Al-Memar, A. Y., Hurles, M. E. et al.(2015). Loss of PCLO function underlies pontocerebellar hypoplasia type III.Neurology 84, 1745.

Ahn, K.-J., Jeong, H. K., Choi, H.-S., Ryoo, S.-R., Kim, Y. J., Goo, J.-S.,Choi, S.-Y., Han, J.-S., Ha, I. and Song, W.-J. (2006). DYRK1A BAC transgenicmice show altered synaptic plasticity with learning and memory defects.Neurobiol. Dis. 22, 463-472.

Altafaj, X., Dierssen, M., Baamonde, C., Martı,́ E., Visa, J., Guimera,̀ J., Oset, M.,González, J. R., Flórez, J., Fillat, C. et al. (2001). Neurodevelopmental delay,motor abnormalities and cognitive deficits in transgenic mice overexpressingDyrk1A (Minibrain), a murine model of Down’s syndrome. Hum. Mol. Genet.10, 1915-1923.

Altafaj, X., Martıń, E. D., Ortiz-Abalia, J., Valderrama, A., Lao-Peregrıń, C.,Dierssen, M. and Fillat, C. (2013). Normalization of Dyrk1A Expression byAAV2/1-shDyrk1A attenuates hippocampal-dependent defects in the Ts65Dnmouse model of Down syndrome. Neurobiol. Dis. 52, 117-127.

Alvarez, M., Altafaj, X., Aranda, S. and de la Luna, S. (2007). DYRK1Aautophosphorylation on serine residue 520 modulates its kinase activity via14-3-3 binding. Mol. Biol. Cell 18, 1167-1178.

Anderson, J. S., Nielsen, J. A., Ferguson, M. A., Burback, M. C., Cox, E. T.,Dai, L., Gerig, G., Edgin, J. O. and Korenberg, J. R. (2013). Abnormal brainsynchrony in Down syndrome. NeuroImage 2, 703-715.

Arron, J. R., Winslow, M. M., Polleri, A., Chang, C.-P., Wu, H., Gao, X.,Neilson, J. R., Chen, L., Heit, J. J., Kim, S. K. et al. (2006). NFAT dysregulationby increased dosage of DSCR1 and DYRK1A on chromosome 21. Nature441, 595-600.

Azorsa, D. O., Robeson, R. L. H., Frost, D., Meec hoovet, B., Brautigam, G. R.,Dickey, C., Beaudry, C., Basu, G. D., Holz, D. R., Hernandez, J. A. et al. (2010).High-content siRNA screening of the kinome identifies kinases involved inAlzheimer’s disease-related tau hyperphosphorylation. BMC Genomics 11, 25.

Ballard, C., Mobley, W., Hardy, J., Williams, G. and Corbett, A. (2016). Dementiain Down’s syndrome. Lancet Neurol. 15, 622-636.

Benfenati, F., Valtorta, F. and Greengard, P. (1991). Computer modeling ofSynapsin I binding to synaptic vesicles and F-actin: implications for regulation ofneurotransmitter release. Proc. Natl Acad. Sci. USA 88, 575-579.

Benfenati, F., Valtorta, F., Rubenstein, J. L., Gorelick, F. S., Greengard, P. andCzernik, A. J. (1992). Synaptic vesicle-associated Ca2+/Calmodulin-dependentprotein kinase II is a binding protein for synapsin I. Nature 359, 417-420.

Branca, C., Shaw, D. M., Belfiore, R., Gokhale, V., Shaw, A. Y., Foley, C., Smith,B., Hulme, C., Dunckley, T., Meechoovet, B. et al. (2017). Dyrk1 inhibitionimproves Alzheimer’s disease-like pathology. Aging Cell. 16, 1146-1154.

Cases-Langhoff, C., Voss, B., Garner, A. M., Appeltauer, U., Takei, K.,Kindler, S., Veh, R. W., De Cmilli, P., Gundelfinger, E. D. and Garner, C. C.(1996). Piccolo, a novel 420 kDa protein associated with the presynapticcytomatrix. Eur. J. Cell Biol. 10, 214-223.

Cesca, F., Baldelli, P., Valtorta, F. and Benfenati, F. (2010). The synapsins: keyactors of synapse function and plasticity. Prog. Neurobiol. 91, 313-348.

Chen, C.-K., Bregere, C., Paluch, J., Lu, J. F., Dickman, D. K. and Chang, K. T.(2014). Activity-dependent facilitation of synaptojanin and synaptic vesiclerecycling by the minibrain kinase. Nat. Commun. 5, ncomms5246.

Courcet, J.-B., Faivre, L., Malzac, P., Masurel-Paulet, A., Lopez, E., Callier, P.,Lambert, L., Lemesle, M., Thevenon, J., Gigot, N. et al. (2012). The DYRK1Agene is a cause of syndromic intellectual disability with severe microcephaly andepilepsy. J. Med. Genet. 49, 731-736.

14


Disea

seModels&Mechan

isms

http://dmm.biologists.org/lookup/doi/10.1242/dmm.035634.supplementalhttp://dmm.biologists.org/lookup/doi/10.1242/dmm.035634.supplementalhttp://dmm.biologists.org/lookup/doi/10.1242/dmm.035634.supplementalhttp://dmm.biologists.org/lookup/doi/10.1242/dmm.035634.supplementalhttp://dmm.biologists.org/lookup/doi/10.1242/dmm.035634.supplementalhttp://dmm.biologists.org/lookup/doi/10.1242/dmm.035634.supplementalhttp://dmm.biologists.org/lookup/doi/10.1242/dmm.035634.supplementalhttp://dmm.biologists.org/lookup/doi/10.1242/dmm.035634.supplementalhttp://dx.doi.org/10.1212/WNL.0000000000001523http://dx.doi.org/10.1212/WNL.0000000000001523http://dx.doi.org/10.1212/WNL.0000000000001523http://dx.doi.org/10.1212/WNL.0000000000001523http://dx.doi.org/10.1016/j.nbd.2005.12.006http://dx.doi.org/10.1016/j.nbd.2005.12.006http://dx.doi.org/10.1016/j.nbd.2005.12.006http://dx.doi.org/10.1016/j.nbd.2005.12.006http://dx.doi.org/10.1093/hmg/10.18.1915http://dx.doi.org/10.1093/hmg/10.18.1915http://dx.doi.org/10.1093/hmg/10.18.1915http://dx.doi.org/10.1093/hmg/10.18.1915http://dx.doi.org/10.1093/hmg/10.18.1915http://dx.doi.org/10.1016/j.nbd.2012.11.017http://dx.doi.org/10.1016/j.nbd.2012.11.017http://dx.doi.org/10.1016/j.nbd.2012.11.017http://dx.doi.org/10.1016/j.nbd.2012.11.017http://dx.doi.org/10.1091/mbc.e06-08-0668http://dx.doi.org/10.1091/mbc.e06-08-0668http://dx.doi.org/10.1091/mbc.e06-08-0668http://dx.doi.org/10.1016/j.nicl.2013.05.006http://dx.doi.org/10.1016/j.nicl.2013.05.006http://dx.doi.org/10.1016/j.nicl.2013.05.006http://dx.doi.org/10.1038/nature04678http://dx.doi.org/10.1038/nature04678http://dx.doi.org/10.1038/nature04678http://dx.doi.org/10.1038/nature04678http://dx.doi.org/10.1186/1471-2164-11-25http://dx.doi.org/10.1186/1471-2164-11-25http://dx.doi.org/10.1186/1471-2164-11-25http://dx.doi.org/10.1186/1471-2164-11-25http://dx.doi.org/10.1016/S1474-4422(16)00063-6http://dx.doi.org/10.1016/S1474-4422(16)00063-6http://dx.doi.org/10.1073/pnas.88.2.575http://dx.doi.org/10.1073/pnas.88.2.575http://dx.doi.org/10.1073/pnas.88.2.575http://dx.doi.org/10.1038/359417a0http://dx.doi.org/10.1038/359417a0http://dx.doi.org/10.1038/359417a0http://dx.doi.org/10.1111/acel.12648http://dx.doi.org/10.1111/acel.12648http://dx.doi.org/10.1111/acel.12648http://dx.doi.org/10.1016/j.pneurobio.2010.04.006http://dx.doi.org/10.1016/j.pneurobio.2010.04.006http://dx.doi.org/10.1038/ncomms5246http://dx.doi.org/10.1038/ncomms5246http://dx.doi.org/10.1038/ncomms5246http://dx.doi.org/10.1136/jmedgenet-2012-101251http://dx.doi.org/10.1136/jmedgenet-2012-101251http://dx.doi.org/10.1136/jmedgenet-2012-101251http://dx.doi.org/10.1136/jmedgenet-2012-101251

Coutadeur, S., Benyamine, H., Delalonde, L., de Oliveira, C., Leblond, B.,Foucourt, A., Besson, T., Casagrande, A.-S., Taverne, T., Girard, A. et al.(2015). A novel DYRK1A (Dual Specificity Tyrosine Phosphorylation-RegulatedKinase 1A) inhibitor for the treatment of Alzheimer’s disease: effect on tau andamyloid pathologies in vitro. J. Neurochem. 133, 440-451.

de la Torre, R., De Sola, S., Pons, M., Duchon, A., de Lagran, M. M., Farré, M.,Fitó, M., Benejam, B., Langohr, K., Rodriguez, J. et al. (2014).Epigallocatechin-3-gallate, a DYRK1A inhibitor, rescues cognitive deficits inDown syndromemousemodels and in humans.Mol. Nutr. FoodRes. 58, 278-288.

de la Torre, R., de Sola, S., Hernandez, G., Farré, M., Pujol, J., Rodriguez, J.,Espadaler, J. M., Langohr, K., Cuenca-Royo, A., Principe, A. et al. (2016).Safety and efficacy of cognitive training plus epigallocatechin-3-gallate inyoung adults with Down’s syndrome (TESDAD): a double-blind, randomised,placebo-controlled, phase 2 trial. Lancet Neurol. 15, 801-810.

Debdab, M., Carreaux, F., Renault, S., Soundararajan, M., Fedorov, O.,Filippakopoulos, P., Lozach, O., Babault, L., Tahtouh, T., Baratte, B. et al.(2011). Leucettines, a class of potent inhibitors of cdc2-like kinases and dualspecificity, tyrosine phosphorylation regulated kinases derived from the marinesponge Leucettamine B: modulation of alternative pre-RNA splicing. J. Med.Chem. 54, 4172-4186.

Di Vona, C., Bezdan, D., Islam, A. B. M. M. K., Salichs, E., López-Bigas, N.,Ossowski, S. and de la Luna, S. (2015). Chromatin-wide profiling of DYRK1Areveals a role as a gene-specific RNA polymerase II CTD kinase. Mol. Cell57, 506-520.

Duchon, A. and Herault, Y. (2016). DYRK1A, a dosage-sensitive gene involved inneurodevelopmental disorders, is a target for drug development in Downsyndrome. Front. Behav. Neurosci. 10, e54285.

Duchon, A., Raveau, M., Chevalier, C., Nalesso, V., Sharp, A. J. and Herault, Y.(2011). Identification of the translocation breakpoints in the Ts65Dn and Ts1Cjemouse lines: relevance for modeling Down syndrome. Mamm. Genome 22, 674.

Ennaceur, A. and Delacour, J. (1988). A new one-trial test for neurobiologicalstudies of memory in rats. 1: behavioral data. Behav. Brain Res. 31, 47-59.

Fassio, A., Patry, L., Congia, S., Onofri, F., Piton, A., Gauthier, J., Pozzi, D.,Messa, M., Defranchi, E., Fadda, M. et al. (2011). SYN1 loss-of-functionmutations in autism and partial epilepsy cause impaired synaptic function. Hum.Mol. Genet. 20, 2297-2307.

Fenster, S. D., Kessels, M. M., Qualmann, B., Chung, W. J., Nash, J.,Gundelfinger, E. D. and Garner, C. C. (2003). Interactions between Piccoloand the actin/dynamin-binding protein Abp1 link vesicle endocytosis topresynaptic active zones. J. Biol. Chem. 278, 20268-20277.

Ferrer, I., Barrachina, M., Puig, B., Martıńez de Lagrán, M., Martı,́ E., Avila J. andDierssen, M. (2005). Constitutive Dyrk1A is abnormally expressed in Alzheimerdisease, Down syndrome, Pick disease, and related transgenic models.Neurobiol. Dis. 20, 392-400.

Freund, T. F., Katona, I. andPiomelli, D. (2003). Role of endogenous cannabinoidsin synaptic signaling. Physiol. Rev. 83, 1017-1066.

Garcia, C. C., Blair, H. J., Seager, M., Coulthard, A., Tennant, S., Buddles, M.,Curtis, A. andGoodship, J. A. (2004). Identification of amutation in Synapsin I, asynaptic vesicle protein, in a family with epilepsy. J. Med. Genet. 41, 183-186.

Garcıá-Cerro, S., Martıńez, P., Vidal, V., Corrales, A., Flórez, J., Vidal, R.,Rueda, N., Arbonés, M. L. and Martıńez-Cué, C. (2014). Overexpressionof Dyrk1A is implicated in several cognitive, electrophysiological andneuromorphological alterations found in a mouse model of Down syndrome.PLoS ONE 9, e106572.

Garner, C. C., Kindler, S. and Gundelfinger, E. D. (2000). Molecular determinantsof presynaptic active zones. Curr. Opin. Neurobiol. 10, 321-327.

Graaf, K. d., Hekerman, P., Spelten, O., Herrmann, A., Packman, L. C.,Büssow, K., Müller-Newen, G. and Becker, W. (2004). Characterization ofCyclin L2, a novel cyclin with an arginine/serine-rich domain phosphorylation byDYRK1A and colocalization with splicing factors. J. Biol. Chem. 279, 4612-4624.

Grau, C., Arató, K., Fernández-Fernández, J. M., Valderrama, A., Sindreu, C.,Fillat, C., Ferrer, I., de la Luna, S. and Altafaj, X. (2014). DYRK1A-mediatedphosphorylation of GluN2A at Ser1048 regulates the surface expression andchannel activity of GluN1/GluN2A receptors. Front. Cell. Neurosci. 8, 331.

Guedj, F., Sébrié, C., Rivals, I., Ledru, A., Paly, E., Bizot, J. C., Smith, D.,Rubin, E., Gillet, B., Arbones, M. et al. (2009). Green tea polyphenols rescue ofbrain defects induced by overexpression of DYRK1A. PLoS ONE 4, e4606.

Guedj, F., Pereira, P. L., Najas, S., Barallobre, M.-J., Chabert, C., Souchet, B.,Sebrie, C., Verney, C., Herault, Y., Arbones, M. et al. (2012). DYRK1A: a masterregulatory protein controlling brain growth. Neurobiol. Dis. 46, 190-203.

Gupta, M., Dhanasekaran, A. R. and Gardiner, K. J. (2016). Mouse models ofDown syndrome: gene content and consequences.Mamm.Genome 27, 538-555.

Hawellek, D. J., Hipp, J. F., Lewis, C. M., Corbetta, M. and Engel, A. K. (2011).Increased functional connectivity indicates the severity of cognitive impairment inmultiple sclerosis. Proc. Natl Acad. Sci. USA 108, 19066-19071.

Head, E., Powell, D., Gold, B. T. and Schmitt, F. A. (2012a). Alzheimer’s disease inDown syndrome. Eur. J. Neurodegener. Dis. 1, 353-364.

Head, E., Silverman, W., Patterson, D. and Lott, I. T. (2012b). Aging and Downsyndrome. Curr. Gerontol. Geriatr. Res. 2012, 412536.

Hilfiker, S., Benfenati, F., Doussau, F., Nairn, A. C., Czernik, A. J.,Augustine, G. J. and Greengard, P. (2005). Structural domains involved in theregulation of transmitter release by synapsins. J. Neurosci. 25, 2658-2669.

Himpel, S., Tegge, W., Frank, R., Leder, S., Joost, H.-G. and Becker, W. (2000).Specificity determinants of substrate recognition by the protein kinase DYRK1A.J. Biol. Chem. 275, 2431-2438.

Hoelter, S. M., Dalke, C., Kallnik, M., Becker, L., Horsch, M., Schrewe, A., Favor,J., Klopstock, T., Beckers, J., Ivandic, B. et al. (2008). “Sighted C3H” mice - atool for analysing the influence of vision on mouse behaviour? Front. Biosci. 13,5810-5823.

Hübner, N. S., Mechling, A. E., Lee, H.-L., Reisert, M., Bienert, T., Hennig, J.,von Elverfeldt, D. and Harsan, L.-A. (2017). The connectomics of braindemyelination: functional and structural patterns in the cuprizone mouse model.Neuroimage 146, 1-18.

Jiang, X., Liu, C., Yu, T., Zhang, L., Meng, K., Xing, Z., Belichenko, P. V.,Kleschevnikov, A. M., Pao, A., Peresie, J. et al. (2015). Genetic dissection of theDown syndrome critical region. Hum. Mol. Genet. 24, 6540-6551.

Jin, N., Yin, X., Gu, J., Zhang, X., Shi, J., Qian, W., Ji, Y., Cao, M., Gu, X., Ding, F.et al. (2015). Truncation and activation of dual specificity tyrosine phosphorylation-regulated kinase 1A by calpain I. J. Biol. Chem. 290, 15219-15237.

Karp, N. A., Meehan, T. F., Morgan, H., Mason, J. C., Blake, A., Kurbatova, N.,Smedley, D., Jacobsen, J., Mott, R. F., Iyer, V. et al. (2015). Applying theARRIVE guidelines to an in vivo database. PLoS Biol. 13, e1002151.

Kawakubo, T., Mori, R., Shirotani, K., Iwata, N. and Asai, M. (2017). NeprilysinIs suppressed by dual-specificity tyrosine-phosphorylation regulated kinase1A (DYRK1A) in Down-syndrome-derived fibroblasts. Biol. Pharm. Bull. 40,327-333.

Khan, N., Afaq, F., Saleem, M., Ahmad, N. and Mukhtar, H. (2006). Targetingmultiple signaling pathways by green tea polyphenol (−)-epigallocatechin-3-gallate. Cancer Res. 66, 2500-2505.

Kilkenny, C., Browne, W. J., Cuthill, I. C., Emerson, M. and Altman, D. G. (2010).Improving bioscience research reporting: the ARRIVE guidelines for reportinganimal research. PLoS Biol. 8, e1000412.

Kim, H., Lee, K.-S., Kim, A.-K., Choi, M., Choi, K., Kang, M., Chi, S.-W.,Lee, M.-S., Lee, J.-S., Lee, S.-Y. et al. (2016). A chemical with proven clinicalsafety rescues down-syndrome-related phenotypes in through DYRK1Ainhibition. Dis. Model. Mech. 9, 839-848.

Kimura, R., Kamino, K., Yamamoto, M., Nuripa, A., Kida, T., Kazui, H.,Hashimoto, R., Tanaka, T., Kudo, T., Yamagata, H. et al. (2007). TheDYRK1A gene, encoded in chromosome 21 Down syndrome critical region,bridges between β-amyloid production and tau phosphorylation in Alzheimerdisease. Hum. Mol. Genet. 16, 15-23.

Korte, G., Dreiseitel, A., Schreier, P., Oehme, A., Locher, S., Geiger, S.,Heilmann, J. and Sand, P. G. (2010). Tea Catechins’ affinity for humancannabinoid receptors. Phytomedicine 17, 19-22.

Leal-Ortiz, S., Waites, C. L., Terry-Lorenzo, R., Zamorano, P.,Gundelfinger, E. D. and Garner, C. C. (2008). Piccolo modulation ofSynapsin1a dynamics regulates synaptic vesicle exocytosis. J. Cell Biol.181, 831.

Li, Z., Yu, T., Morishima, M., Pao, A., LaDuca, J., Conroy, J., Nowak, N.,Matsui, S.-I., Shiraishi, I. and Yu, Y. E. (2007). Duplication of the entire22.9 Mb human chromosome 21 syntenic region on mouse chromosome 16causes cardiovascular and gastrointestinal abnormalities. Hum. Mol. Genet.16, 1359-1366.

Litovchick, L., Florens, L. A., Swanson, S. K., Washburn, M. P. andDeCaprio, J. A. (2011). DYRK1A protein kinase promotes quiescence andsenescence through DREAM complex assembly. Genes Dev. 25, 801-813.

Llinás, R., McGuinness, T. L., Leonard, C. S., Sugimori, M. and Greengard, P.(1985). Intraterminal injection of Synapsin I or Calcium/Calmodulin-dependentprotein kinase II alters neurotransmitter release at the squid giant synapse. Proc.Natl. Acad. Sci. USA 82, 3035.

Llinás, R., Gruner, J. A., Sugimori, M., McGuinness, T. L. and Greengard, P.(1991). Regulation by Synapsin I and Ca(2+)-Calmodulin-dependent proteinkinase II of the transmitter release in squid giant synapse. J. Physiol. 436, 257.

Mao, J., Maye, P., Kogerman, P., Tejedor, F. J., Toftgard, R., Xie, W., Wu, G. andWu, D. (2002). Regulation of Gli1 transcriptional activity in the nucleus by Dyrk1.J. Biol. Chem. 277, 35156-35161.

Marchal, J. P., Maurice-Stam, H., Houtzager, B. A., Rutgersvan Rozenburg-Marres, S. L., Oostrom, K. J., Grootenhuis, M. A. andvan Trotsenburg, A. S. P. (2016). Growing up with Down syndrome:development from 6 months to 10.7 years. Res. Dev. Disabil. 59, 437-450.

Martı,́ E., Altafaj, X., Dierssen, M., de la Luna, S., Fotaki, V., Alvarez, M.,Pérez-Riba, M., Ferrer, I. and Estivill, X. (2003). Dyrk1A expression patternsupports specific roles of this kinase in the adult central nervous system. BrainRes. 964, 250-263.

Martıńez de Lagrán, M., Altafaj, X., Gallego, X., Martı,́ E., Estivill, X., Sahún, I.,Fillat, C. and Dierssen, M. (2004). Motor phenotypic alterations in TgDyrk1atransgenic mice implicate DYRK1A in Down syndrome motor dysfunction.Neurobiol. Dis. 15, 132-142.

15


Disea

seModels&Mechan

isms

http://dx.doi.org/10.1111/jnc.13018http://dx.doi.org/10.1111/jnc.13018http://dx.doi.org/10.1111/jnc.13018http://dx.doi.org/10.1111/jnc.13018http://dx.doi.org/10.1111/jnc.13018http://dx.doi.org/10.1002/mnfr.201300325http://dx.doi.org/10.1002/mnfr.201300325http://dx.doi.org/10.1002/mnfr.201300325http://dx.doi.org/10.1002/mnfr.201300325http://dx.doi.org/10.1016/S1474-4422(16)30034-5http://dx.doi.org/10.1016/S1474-4422(16)30034-5http://dx.doi.org/10.1016/S1474-4422(16)30034-5http://dx.doi.org/10.1016/S1474-4422(16)30034-5http://dx.doi.org/10.1016/S1474-4422(16)30034-5http://dx.doi.org/10.1021/jm200274dhttp://dx.doi.org/10.1021/jm200274dhttp://dx.doi.org/10.1021/jm200274dhttp://dx.doi.org/10.1021/jm200274dhttp://dx.doi.org/10.1021/jm200274dhttp://dx.doi.org/10.1021/jm200274dhttp://dx.doi.org/10.1016/j.molcel.2014.12.026http://dx.doi.org/10.1016/j.molcel.2014.12.026http://dx.doi.org/10.1016/j.molcel.2014.12.026http://dx.doi.org/10.1016/j.molcel.2014.12.026http://dx.doi.org/10.3389/fnbeh.2016.00104http://dx.doi.org/10.3389/fnbeh.2016.00104http://dx.doi.org/10.3389/fnbeh.2016.00104http://dx.doi.org/10.1007/s00335-011-9356-0http://dx.doi.org/10.1007/s00335-011-9356-0http://dx.doi.org/10.1007/s00335-011-9356-0http://dx.doi.org/10.1016/0166-4328(88)90157-Xhttp://dx.doi.org/10.1016/0166-4328(88)90157-Xhttp://dx.doi.org/10.1093/hmg/ddr122http://dx.doi.org/10.1093/hmg/ddr122http://dx.doi.org/10.1093/hmg/ddr122http://dx.doi.org/10.1093/hmg/ddr122http://dx.doi.org/10.1074/jbc.M210792200http://dx.doi.org/10.1074/jbc.M210792200http://dx.doi.org/10.1074/jbc.M210792200http://dx.doi.org/10.1074/jbc.M210792200http://dx.doi.org/10.1016/j.nbd.2005.03.020http://dx.doi.org/10.1016/j.nbd.2005.03.020http://dx.doi.org/10.1016/j.nbd.2005.03.020http://dx.doi.org/10.1016/j.nbd.2005.03.020http://dx.doi.org/10.1152/physrev.00004.2003http://dx.doi.org/10.1152/physrev.00004.2003http://dx.doi.org/10.1136/jmg.2003.013680http://dx.doi.org/10.1136/jmg.2003.013680http://dx.doi.org/10.1136/jmg.2003.013680http://dx.doi.org/10.1371/journal.pone.0106572http://dx.doi.org/10.1371/journal.pone.0106572http://dx.doi.org/10.1371/journal.pone.0106572http://dx.doi.org/10.1371/journal.pone.0106572http://dx.doi.org/10.1371/journal.pone.0106572http://dx.doi.org/10.1016/S0959-4388(00)00093-3http://dx.doi.org/10.1016/S0959-4388(00)00093-3http://dx.doi.org/10.1074/jbc.M310794200http://dx.doi.org/10.1074/jbc.M310794200http://dx.doi.org/10.1074/jbc.M310794200http://dx.doi.org/10.1074/jbc.M310794200http://dx.doi.org/10.3389/fncel.2014.00331http://dx.doi.org/10.3389/fncel.2014.00331http://dx.doi.org/10.3389/fncel.2014.00331http://dx.doi.org/10.3389/fncel.2014.00331http://dx.doi.org/10.1371/journal.pone.0004606http://dx.doi.org/10.1371/journal.pone.0004606http://dx.doi.org/10.1371/journal.pone.0004606http://dx.doi.org/10.1016/j.nbd.2012.01.007http://dx.doi.org/10.1016/j.nbd.2012.01.007http://dx.doi.org/10.1016/j.nbd.2012.01.007http://dx.doi.org/10.1007/s00335-016-9661-8http://dx.doi.org/10.1007/s00335-016-9661-8http://dx.doi.org/10.1073/pnas.1110024108http://dx.doi.org/10.1073/pnas.1110024108http://dx.doi.org/10.1073/pnas.1110024108http://dx.doi.org/10.1155/2012/412536http://dx.doi.org/10.1155/2012/412536http://dx.doi.org/10.1523/JNEUROSCI.4278-04.2005http://dx.doi.org/10.1523/JNEUROSCI.4278-04.2005http://dx.doi.org/10.1523/JNEUROSCI.4278-04.2005http://dx.doi.org/10.1074/jbc.275.4.2431http://dx.doi.org/10.1074/jbc.275.4.2431http://dx.doi.org/10.1074/jbc.275.4.2431http://dx.doi.org/10.1016/j.neuroimage.2016.11.008http://dx.doi.org/10.1016/j.neuroimage.2016.11.008http://dx.doi.org/10.1016/j.neuroimage.2016.11.008http://dx.doi.org/10.1016/j.neuroimage.2016.11.008http://dx.doi.org/10.1093/hmg/ddv364http://dx.doi.org/10.1093/hmg/ddv364http://dx.doi.org/10.1093/hmg/ddv364http://dx.doi.org/10.1074/jbc.M115.645507http://dx.doi.org/10.1074/jbc.M115.645507http://dx.doi.org/10.1074/jbc.M115.645507http://dx.doi.org/10.1371/journal.pbio.1002151http://dx.doi.org/10.1371/journal.pbio.1002151http://dx.doi.org/10.1371/journal.pbio.1002151http://dx.doi.org/10.1248/bpb.b16-00825http://dx.doi.org/10.1248/bpb.b16-00825http://dx.doi.org/10.1248/bpb.b16-00825http://dx.doi.org/10.1248/bpb.b16-00825http://dx.doi.org/10.1158/0008-5472.CAN-05-3636http://dx.doi.org/10.1158/0008-5472.CAN-05-3636http://dx.doi.org/10.1158/0008-5472.CAN-05-3636http://dx.doi.org/10.1371/journal.pbio.1000412http://dx.doi.org/10.1371/journal.pbio.1000412http://dx.doi.org/10.1371/journal.pbio.1000412http://dx.doi.org/10.1242/dmm.025668http://dx.doi.org/10.1242/dmm.025668http://dx.doi.org/10.1242/dmm.025668http://dx.doi.org/10.1242/dmm.025668http://dx.doi.org/10.1093/hmg/ddl437http://dx.doi.org/10.1093/hmg/ddl437http://dx.doi.org/10.1093/hmg/ddl437http://dx.doi.org/10.1093/hmg/ddl437http://dx.doi.org/10.1093/hmg/ddl437http://dx.doi.org/10.1016/j.phymed.2009.10.001http://dx.doi.org/10.1016/j.phymed.2009.10.001http://dx.doi.org/10.1016/j.phymed.2009.10.001http://dx.doi.org/10.1083/jcb.200711167http://dx.doi.org/10.1083/jcb.200711167http://dx.doi.org/10.1083/jcb.200711167http://dx.doi.org/10.1083/jcb.200711167http://dx.doi.org/10.1093/hmg/ddm086http://dx.doi.org/10.1093/hmg/ddm086http://dx.doi.org/10.1093/hmg/ddm086http://dx.doi.org/10.1093/hmg/ddm086http://dx.doi.org/10.1093/hmg/ddm086http://dx.doi.org/10.1101/gad.2034211http://dx.doi.org/10.1101/gad.2034211http://dx.doi.org/10.1101/gad.2034211http://dx.doi.org/10.1073/pnas.82.9.3035http://dx.doi.org/10.1073/pnas.82.9.3035http://dx.doi.org/10.1073/pnas.82.9.3035http://dx.doi.org/10.1073/pnas.82.9.3035http://dx.doi.org/10.1113/jphysiol.1991.sp018549http://dx.doi.org/10.1113/jphysiol.1991.sp018549http://dx.doi.org/10.1113/jphysiol.1991.sp018549http://dx.doi.org/10.1074

Correction of cognitive deficits in mouse models of Down ...1Institut de Géne ́tique et de...

Documents

Transcript of Correction of cognitive deficits in mouse models of Down ...1Institut de Géne ́tique et de...